[70035] trunk/dports/graphics/ImageMagick/Portfile

[Ryan Schmidt]

On Jul 27, 2010, at 12:04, Eric Hall wrote:

> 	Given that p5-perlmagick requires ImageMagick to be installed, why have
> a separate port for it (other than the fact that MP doesn't yet have 
> variants-in-dependencies)?

That's one reason: it lets other ports declare a dependency on it -- today -- without waiting for a resolution to #126, which may or may not ever materialize.

The other reason is that it's preferable to have separate features in separate ports, so that they can be installed separately without having to rebuild the main part of the software. ImageMagick takes a bit of time to build; it's nice to be able to later add perl functionality without having to rebuild all of ImageMagick. That's why I spent a lot of effort splitting variants out of the php5 port into separate ports last year in #19091.


> 	As well, saying "hopefully it even works" doesn't give a lot of confidence
> that p5-perlmagick is ready for anyone to use it, while I know quite clearly
> that ImageMagick +perl works rather well (we use it on a regular basis).

Ok. I personally don't know whether either of them work as I've never used them. Unfortunately, with the volume of updates I commit, the most I generally do is verify something builds on at least one Mac OS X system before I commit.


> 	If the perl interface really has to be pulled out, I think the latter 
> (have p5-perlmagick use the ImageMagick perl source) is the correct way to do it.
> Otherwise (given the history), there's a continual mismatch between the
> ImageMagick version and the p5-perlmagick version, just asking for oddball
> issues to crop up.
> 	How about making a ticket with your efforts to have p5-perlmagick use
> the perlmagick source inside of the ImageMagick source code and putting back
> ImageMagick +perl until that effort can be completed?

That sounds like a sensible idea; I should have done that in the first place. I put the +perl variant back for now and filed

http://trac.macports.org/ticket/25844